From: Yicong Yang <yangyicong@hisilicon.com>
To: Bjorn Helgaas <helgaas@kernel.org>
Cc: <bhelgaas@google.com>, <linux-pci@vger.kernel.org>,
<linuxarm@huawei.com>,
Mika Westerberg <mika.westerberg@linux.intel.com>,
"Rafael J. Wysocki" <rafael.j.wysocki@intel.com>,
Peter Wu <peter@lekensteyn.nl>
Subject: Re: [PATCH] PCI: Make sure the bus bridge powered on when scanning bus
Date: Fri, 18 Sep 2020 17:31:54 +0800 [thread overview]
Message-ID: <3fc0ea97-d0ed-22ad-5906-8d9e98920ffd@hisilicon.com> (raw)
In-Reply-To: <20200917210737.GA1732082@bjorn-Precision-5520>
On 2020/9/18 5:07, Bjorn Helgaas wrote:
> [+cc Mika, Rafael, Peter]
>
> On Wed, Jul 29, 2020 at 07:30:23PM +0800, Yicong Yang wrote:
>> When the bus bridge is runtime suspended, we'll fail to rescan
>> the devices through sysfs as we cannot access the configuration
>> space correctly when the bridge is in D3hot.
>> It can be reproduced like:
>>
>> $ echo 1 > /sys/bus/pci/devices/0000:80:00.0/0000:81:00.1/remove
>> $ echo 1 > /sys/bus/pci/devices/0000:80:00.0/pci_bus/0000:81/rescan
>>
>> 0000:80:00.0 is root port and is runtime suspended and we cannot
>> get 0000:81:00.1 after rescan.
>>
>> Make bridge powered on when scanning the child bus, by adding
>> pm_runtime_get_sync()/pm_runtime_put() in pci_scan_child_bus_extend().
>>
>> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
>> ---
>> drivers/pci/probe.c | 11 +++++++++++
>> 1 file changed, 11 insertions(+)
>>
>> diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
>> index 2f66988..5bb502b 100644
>> --- a/drivers/pci/probe.c
>> +++ b/drivers/pci/probe.c
>> @@ -2795,6 +2795,14 @@ static unsigned int pci_scan_child_bus_extend(struct pci_bus *bus,
>>
>> dev_dbg(&bus->dev, "scanning bus\n");
>>
>> + /*
>> + * Make sure the bus bridge is powered on, otherwise we may not be
>> + * able to scan the devices as we may fail to access the configuration
>> + * space of subordinates.
>> + */
>> + if (bus->self)
>> + pm_runtime_get_sync(&bus->self->dev);
> I think if we do this, we should be able to remove the call from
> pci_scan_bridge() added by d963f6512e15 ("PCI: Power on bridges before
> scanning new devices"), right?
>
> The reason we need it here is because there are two paths to
> pci_scan_child_bus_extend() and only one of them calls
> pm_runtime_get_sync():
>
> pci_scan_bridge_extend
> pm_runtime_get_sync
> pci_scan_child_bus_extend
>
> pci_scan_child_bus
> pci_scan_child_bus_extend
>
> If we move the pm_runtime_get_sync() from pci_scan_bridge_extend() to
> pci_scan_child_bus_extend(), both paths should be safe.
A bit different, I think. The issue I met is a bit different from Mika, as
we go through different sysfs files. Think about rescanning device under a
root port,
when echo 1 > /sysfs/bus/pci/devices/${RootPort}/rescan:
rescan_restore()
pci_rescan_bus(pdev->bus) /* we will rescan the root bus */
pci_rescan_child_bus()
pci_scan_child_bus_extend() /* we cannot wake up the bus bridge here as is on the root bus */
pci_scan_bridge_extend() /* we have to wake up the root port here */
when echo 1 > /sysfs/bus/pci/devices/${RootPort}/pci_bus/${PciBus}/rescan:
rescan_restore()
pci_rescan_bus(bus) /* we will rescan the bus of the root port */
pci_rescan_child_bus()
pci_scan_child_bus_extend() /* we can wake up the bus bridge - root port here */
As different bus is rescanned, so it'll have problem without patch d963f6512e15.
>
>> /* Go find them, Rover! */
>> for (devfn = 0; devfn < 256; devfn += 8) {
>> nr_devs = pci_scan_slot(bus, devfn);
>> @@ -2907,6 +2915,9 @@ static unsigned int pci_scan_child_bus_extend(struct pci_bus *bus,
>> }
>> }
>>
>> + if (bus->self)
>> + pm_runtime_put(&bus->self->dev);
> I would probably do this:
>
> struct pci_dev *bridge = bus->self;
>
> if (bridge)
> pm_runtime_get_sync(&bridge->dev);
> ...
> if (bridge)
> pm_runtime_put(&bridge->dev);
Sure.
Regards,
Yicong
>
>> /*
>> * We've scanned the bus and so we know all about what's on
>> * the other side of any bridges that may be on this bus plus
>> --
>> 2.8.1
>>
> .
>
next prev parent reply other threads:[~2020-09-18 9:32 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-29 11:30 [PATCH] PCI: Make sure the bus bridge powered on when scanning bus Yicong Yang
2020-08-21 9:54 ` Yicong Yang
2020-09-17 21:07 ` Bjorn Helgaas
2020-09-18 9:31 ` Yicong Yang [this message]
2020-09-18 16:17 ` Bjorn Helgaas
2020-09-19 10:22 ` Yicong Yang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3fc0ea97-d0ed-22ad-5906-8d9e98920ffd@hisilicon.com \
--to=yangyicong@hisilicon.com \
--cc=bhelgaas@google.com \
--cc=helgaas@kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linuxarm@huawei.com \
--cc=mika.westerberg@linux.intel.com \
--cc=peter@lekensteyn.nl \
--cc=rafael.j.wysocki@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).